# Representation Forcing for Bottleneck-Free Unified Multimodal Models

Page: https://stenobird.com/podcast/daily-paper-cast-7079649/representation-forcing-for-bottleneck-free-unified-multimodal-models
Text version: https://stenobird.com/podcast/daily-paper-cast-7079649/representation-forcing-for-bottleneck-free-unified-multimodal-models.md
Podcast: [Daily Paper Cast](https://stenobird.com/podcast/daily-paper-cast-7079649)
Published: 2026-06-02T04:14:34+00:00
Episode link: https://share.transistor.fm/s/fa685904
Audio file: https://media.transistor.fm/fa685904/d30cf5de.mp3
Processing state: not_requested
JSON: https://stenobird.com/v1/public/podcasts/daily-paper-cast-7079649/episodes/representation-forcing-for-bottleneck-free-unified-multimodal-models
Duration seconds: 1466

## Resource

🤗 Upvotes: 44 | cs.CV Authors: Yuqing Wang, Zhijie Lin, Ceyuan Yang, Yang Zhao, Fei Xiao, Hao He, Qi Zhao, Zihan Ding, Fuyun Wang, Shuai Wang, Youliang Zhang, Haoqi Fan, Xihui Liu Title: Representation Forcing for Bottleneck-Free Unified Multimodal Models Arxiv: http://arxiv.org/abs/2605.31604v1 Abstract: Unified multimodal models (UMMs) aim to handle perception and generation in a single model. Yet existing UMMs still rely on a frozen, separately pretrained VAE for image generation, imposing a structural bottleneck. Naively removing it introduces a quality gap, as the model must learn both high-level structure and low-level details from raw pixels. In this paper, we propose Representation Forcing (RF), a technique that closes this gap by making representation prediction a native capability of the model. Concretely, RF forces the decoder to autoregressively predict visual representations as intermediate tokens before pixels; these tokens then stay in context to guide pixel diffusion within the same backbone. By turning representations from perception outputs into generation targets, RF eliminates the need for any external generative latent space. We find that RF benefits both understanding and generation. On image generation, our pixel-space model with RF matches state-of-the-art VAE-based unified models. On image understanding, pixel-space RF generally outperforms its VAE-based variant. Together, these results offer an effective step toward end-to-end, bottleneck-free UMMs.

## Actions

- request_transcript: `POST https://stenobird.com/v1/public/podcasts/daily-paper-cast-7079649/episodes/representation-forcing-for-bottleneck-free-unified-multimodal-models/transcription-requests` — Idempotently request low-priority transcript generation for this episode.
- read_markdown: `GET https://stenobird.com/podcast/daily-paper-cast-7079649/representation-forcing-for-bottleneck-free-unified-multimodal-models.md` — Read the agent-friendly Markdown representation of this episode resource.

A page view does not enqueue transcription. Agents should invoke `request_transcript` explicitly when they need this episode processed.

## Transcript

Full transcripts are not published on public pages unless there is a clear rights basis.